Merge pull request #2 from mazingstudio/dev

Handle more cases

Author: ZuraGuerra

README.md

@@ -31,6 +31,12 @@ Norma depends heavily on the implementation of the `URI` module (part of the sta
 iex> Norma.normalize("mazing.studio")
 > "http://mazing.studio"
 
+iex> Norma.normalize_if_valid("mazing.studio")
+> {:ok, "http://mazing.studio"}
+
+iex> Norma.normalize_if_valid("mazing")
+> {:error, "Not an URL."}
+
 iex> options = %{remove_fragment: true, force_root_path: true, remove_www: true}
 iex> Norma.normalize("//www.mazing.studio:1337/test#test", options)
 > "http://mazing.studio/"

lib/norma.ex

@@ -15,12 +15,66 @@ defmodule Norma do
   ## Examples
 
       iex> Norma.normalize("//www.mazing.studio", %{remove_www: true})
-      "http://mazing.studio"
+      {:ok, "http://mazing.studio"}
 
   """
+  def normalize_if_valid(url, options \\ %{}) do
+    if is_url?(url) do
+      normalized_url = url
+        |> safe_parse
+        |> Normalizer.normalize(options)
+      {:ok, normalized_url}
+    else
+      {:error, "Not an URL."}
+    end
+  end
+
+  @doc """
+  Similar to `normalize_if_valid/2`, but will return the given string unchanged
+  if it's not an URL.
+
+  ## Examples
+
+      iex> Norma.normalize!("//www.mazing.studio", %{remove_www: true})
+      "http://mazing.studio"
+  """
   def normalize(url, options \\ %{}) do
+    case normalize_if_valid(url, options) do
+      {:ok, normalized_url} -> normalized_url
+      {:error, _} -> url
+    end
+  end
+
+  @doc """
+  Solve an issue related to the regex provided by the URI spec
+  (see https://tools.ietf.org/html/rfc3986#appendix-B).
+
+  If trying to parse from string to %URI{} something like "mazing.studio:80",
+  the result will be:
+    %URI{scheme: "mazing.studio", path: "21", host: nil}
+    _(Other keys skipped for brevity, but their value is `nil`.)_
+
+  But "//mazing.studio:80", will be parsed correctly:
+    %URI{host: "mazing.studio", authority: "mazing.studio:80", port: 80}
+  """
+  defp safe_parse(url) do
     url
     |> URI.parse
-    |> Normalizer.normalize(options)
+    |> has_valid_host?
   end
+
+  defp has_valid_host?(url = %URI{host: nil}) do
+    url = url |> URI.to_string
+
+    "//" <> url
+    |> safe_parse
+  end
+
+  defp has_valid_host?(url = %URI{host: host})
+  when host != nil, do: url
+
+  @doc """
+  This sure looks dumb, but a valid host will normally have at least a dot.
+  """
+  defp is_url?(url), do: url |> String.contains?(".")
 end

lib/norma/utils.ex

@@ -1,5 +1,7 @@
 defmodule Norma.Utils do
-  def port_handler (port) do
+  @masked_ports ["443", "80", "8080", "21"]
+
+  def port_handler(port) do
     case port do
       443  -> "https"
       80   -> "http"
@@ -24,15 +26,17 @@ defmodule Norma.Utils do
   Help would be appreciated here to solve it better.
   """
   def form_url(%URI{host: host,
-                     path: path,
-                     scheme: scheme,
-                     query: query,
-                     fragment: fragment}) do
-    form_scheme(scheme)                   # "http://"
-    |> safe_concat(host)                  # "http://google.com"
-    |> safe_concat(path)                  # "http://google.com/test"
-    |> Kernel.<>(form_fragment(fragment)) # "http://google.com/test#cats"
-    |> Kernel.<>(form_query(query))       # "http://google.com/test#cats?dogs_allowed=ño"
+                    path: path,
+                    scheme: scheme,
+                    query: query,
+                    port: port,
+                    fragment: fragment}) do
+    form_scheme(scheme)                         # "http://"
+    |> safe_concat(host)                        # "http://google.com"
+    |> Kernel.<>(form_port(port |> to_string))  # "http://google.com:1337"
+    |> safe_concat(path)                        # "http://google.com:1337/test"
+    |> Kernel.<>(form_fragment(fragment))       # "http://google.com:1337/test#cats"
+    |> Kernel.<>(form_query(query))             # "http://google.com:1337/test#cats?dogs_allowed=ño"
   end
 
   defp form_scheme(""), do: ""
@@ -44,6 +48,11 @@ defmodule Norma.Utils do
   defp form_query(nil), do: ""
   defp form_query(query), do: "?" <> query
 
+  defp form_port(""), do: ""
+  defp form_port(port)
+  when port in @masked_ports, do: ""
+  defp form_port(port), do: ":" <> port
+
   defp safe_concat(left, right) do
     left  = left  || ""
     right = right || ""

mix.exs

@@ -4,8 +4,8 @@ defmodule Norma.Mixfile do
   def project do
     [
       app: :norma,
-      version: "1.2.0",
-      elixir: "~> 1.2",
+      version: "1.4.2",
+      elixir: "~> 1.3",
       start_permanent: Mix.env == :prod,
       description: description(),
       package: package(),

test/norma_test.exs

@@ -1,15 +1,23 @@
 defmodule NormaTest do
   use ExUnit.Case
+
   @with_scheme "https://mazing.studio"
   @without_scheme "mazing.studio"
-  @without_scheme_but_port "//mazing.studio:21"
+  @without_scheme_but_path "mazing.studio/test"
+  @without_scheme_but_port "mazing.studio:21"
+  @without_scheme_but_port_alt "mazing.studio:1337"
+  @without_scheme_but_port_and_path "mazing.studio:1337/test"
   @with_path "https://mazing.studio/test"
   @with_fragment "https://mazing.studio#test"
   @with_www "https://www.mazing.studio"
   @full_example "//www.mazing.studio:1337/test#test"
 
-  test "scheme defaults to `http` when not provided (nor a port)",
-    do: assert Norma.normalize(@without_scheme) == "http://mazing.studio"
+  test "scheme defaults to `http` when not provided (nor a port)" do
+    assert Norma.normalize(@without_scheme) == "http://mazing.studio"
+    assert Norma.normalize(@without_scheme_but_path) == "http://mazing.studio/test"
+    assert Norma.normalize(@without_scheme_but_port_alt) == "http://mazing.studio:1337"
+    assert Norma.normalize(@without_scheme_but_port_and_path) == "http://mazing.studio:1337/test"
+  end
 
   test "scheme gets infered from port",
     do: assert Norma.normalize(@without_scheme_but_port) == "ftp://mazing.studio"
@@ -34,6 +42,9 @@ defmodule NormaTest do
     do: assert Norma.normalize(@full_example,
                                %{force_root_path: true,
                                  remove_fragment: true,
-                                      remove_www: true}) == "http://mazing.studio/"
+                                      remove_www: true}) == "http://mazing.studio:1337/"
+
+  test "identify invalid URL",
+    do: assert Norma.normalize_if_valid("mazing") == {:error, "Not an URL."}
 
 end